Mike Paulin*

Department of Zoology and Centre for Neuroscience, University of Otago, Dunedin, New Zealand

Coding of Dynamic Information by Vestibular Neurons

Using receptive fields it is possible to understand at an elementary level how spatial aspects of a stimulus are encoded by neurons, and how that information may be used in neural computations. Understanding dynamical aspects of neural behaviour would seem to require differential equations or transfer functions, which are in general more difficult to deal with than receptive fields. The idea of a receptive field can be generalized in a way that retains conceptual and computational elegance while acquiring the quantitative features of differential equations. We illustrate this using generalized receptive fields of bullfrog vestibular neurons. These receptive fields are constructed in a mathematical space whose axes are the state variables of the equation of motion of the bullfrog's head. Receptive fields of vestibular sensory neurons form a map in this space. The state space receptive field of a given neuron is the probability density of head state conditional upon the occurrence of a spike. Single spikes can be interpreted as packets of information about the dynamical state of the head. Computation in assemblies of spiking neurons can be understood in terms of exchanging, transforming and combining these packets.

* And Larry Hoffman, UCLA School of Medicine and NASA-Ames Vestibular Research Facility

This page was last updated on 16 Feb 2001.

Back.